I’m on a trial run.
I successfully installed and ran a germline pipeline on AWS instance p3.8xlarge.
Next, we decided to run the pipeline on the biggest instance out there (p4d.24xlarge).
Installation seemed to have been successful. But when I start germline pipeline, it ended up with errors.
The command I used:
pbrun germline --ref /mnt/d1/Ref/Homo_sapiens_assembly38.fasta --in-fq /mnt/d1/PD01/FQ/H1_S120_L003_R1_001.fastq.gz /mnt/d1/PD01/FQ/H1_S120_L003_R2_001.fastq.gz --knownSites /mnt/d1/Ref/Homo_sapiens_assembly38.dbsnp138.vcf --knownSites /mnt/d1/Ref/Mills_and_1000G_gold_standard.indels.hg38.vcf --out-bam /mnt/d1/PD01/germline/H1.parabricks.bam --out-variants /mnt/d1/PD01/germline/H1.parabricks.g.vcf.gz --out-recal-file /mnt/d1/PD01/germline/H1.pararabricks_recal.txt --read-group-sm H1 --read-group-lb BCH --read-group-pl ILLUMINA --gvcf --num-gpus 8 --tmp-dir /mnt/d2/tmp/ --keep-tmp
Error message I see:
GPU-BWA mem
ProgressMeter Reads Base Pairs Aligned
cudaSafeCall() failed at ParaBricks/src/mem_chain_kernel.cu:148 : invalid device symbol
cudaSafeCall() failed at ParaBricks/src/mem_chain_kernel.cu:148 : invalid device symbol
cudaSafeCall() failed at ParaBricks/src/mem_chain_kernel.cu:148 : invalid device symbol
cudaSafeCall() failed at ParaBricks/src/mem_chain_kernel.cu:148 : invalid device symbol
cudaSafeCall() failed at ParaBricks/src/mem_chain_kernel.cu:148 : invalid device symbol
cudaSafeCall() failed at ParaBricks/src/mem_chain_kernel.cu:148 : invalid device symbol
cudaSafeCall() failed at ParaBricks/src/mem_chain_kernel.cu:148 : invalid device symbol
cudaSafeCall() failed at ParaBricks/src/mem_chain_kernel.cu:148 : invalid device symbol
Please contact Parabricks-Support@nvidia.com for any questions
Is it a license issue? Or am I missing some setting or library for GPU/Cuda?
Thank you